Previous abstract Next abstract
A simple decision tree scheme has been devised to classify objects on Palomar Observatory Sky Survey~II $J$ plates. Objects are classified as stars or nonstars; nonstellar objects may be galaxies or artifacts. It is straightforward to generalize the algorithm to more classes. The decision tree has two tiers on which objects are separated; separation is done across a linear boundary in a space spanned by two image parameters. The distinctive and self-similar nature of stellar images makes such a simple classifier possible. Nearly 1,400 stars and 1,300 galaxies in the vicinity of the Coma cluster identified from the literature and by eye were used to train the classifier. A preliminary selection of stars for the UJ1.0 catalogue is made using the classifier until deeper plate material is available for corroborative stellar identification. This paper will discuss the classifier and its implementation in the first version of the USNO Short-J catalog UJ1.0.
Monday program listing